Engine Works

Under the hood of Alteryx: tips, tricks and how-tos.
DanH
Alteryx
Alteryx

Alteryx Architectures_Banner_999x240-02.png

 

 

Welcome to the second article in the Alteryx Architectures blog series.  In this installment, well take a look at what we call “starter architectures,”   common architectures often seen with initial installations of Alteryx Server.  These relatively simple designs can then be scaled in the future to accommodate evolving performance and management needs. 

 

Introduction

When architecting an Alteryx Server environment, in the best case scenario, we have data from an existing workload to analyze and then guide a design. However, if you are installing Alteryx Server for the first time, you may not have much useful execution data to analyze.

 

Without historic execution data from an existing Alteryx Server, we typically estimate the expected workload, and focus on one of the following designs as a starting point. After you deploy and gain experience operating within a starter architecture, you’ll also start generating some useful execution and performance metrics. That data and experience can then guide scaling and optimization beyond the initial starter architecture.

 

Alteryx technical associates can help you in both the initial design of your environment, or when scaling and optimizing your existing architecture. The following designs are not a substitute for a sizing exercise to determine the appropriately sized Alteryx Server deployment to best fit your organization's business requirements, workload patterns, and data sizes.  The following designs are not necessarily the bare minimum hardware/licensing requirement as noted here, but rather starter designs which are most commonly recommended. These recommendations are derived from Alteryx internal benchmarking as well as observations from customer experiences.

 

Starter Architecture A

  Single-node Alteryx Server

Starter Architecture A- Single node with Controller, Worker, Database, and Gallery services.Starter Architecture A- Single node with Controller, Worker, Database, and Gallery services.

 

Specifications

  • Alteryx Server, single-node
    • All four services configured: Controller, Worker, Gallery, and MongoDB Database
      • Worker configured to process four simultaneous workflows
    • 8 CPU Cores (physical)
    • 128 GB RAM
    • Disk:
      • C : 100GB(+)
      • D : 1TB(+) SSD with installed Location/Business datasets, or 500GB(+) SSD without

Capabilities

  • Standard Server Functionality
    • Ad-hoc, web-based workflow execution
    • Scheduled workflow execution
    • Centralized workflow version control
    • REST API for execution, administration, and audit
    • Centralized management of authorization (for execution, data connections, credentials)
  • Capable of providing 240 minutes of workflow execution per hour *
  • Capable of serving up to 100 concurrent web user sessions *

Alteryx Licensing

Alteryx Server uses a core-based licensing model. The following licenses will serve the operation of the Alteryx Server architecture described above:

  • 1 x Alteryx Server
  • 2 x Alteryx Server 2-Core Additional Capacity

Environment Summary

In this architecture, all of the four main service components of Server (Controller, Worker, MongoDB persistence, and the Gallery front-end) reside on a single node. When we platform all four services on a single node, we generally recommend eight CPU cores as a starting point, which allows for four simultaneously processing workflows (using classic E1 engine).

 

* Based on average behavior and recommended configurations. Your performance may vary.

 

Starter Architecture B

 Two-node Environment

two.PNG

 

Specifications

  • Primary node
    • Multiple services configured: Controller, Worker, Gallery, and MongoDB Database
      • Worker configured to process four simultaneous workflows
    • 8 CPU Cores (physical)
    • 128GB RAM
    • Disk:
      • C : 100GB(+)
      • D : 1TB(+) SSD with installed Location/Business datasets, or 500GB(+) SSD without
  • Worker node
    • 8 CPU Cores (physical)
    • 128GB RAM
    • Disk:
      • C : 100GB(+)
      • D : 1TB(+) SSD with installed Location/Business datasets, or 500GB(+) SSD without
    • Worker service configured
      • Worker configured to process four simultaneous workflows

Capabilities

  • Standard Server Functionality
    • Ad-hoc, web-based workflow execution
    • Scheduled workflow execution
    • Centralized workflow version control
    • REST API for execution, administration, and audit
    • Centralized management of authorization (for execution, data connections, credentials)
  • Capable of providing 480 minutes of workflow execution per hour *
  • Capable of serving up to 100 concurrent web user sessions *
  • Workload management options, including Worker assignment

 

Alteryx Licensing

Alteryx Server uses a core-based licensing model. The following licenses will serve the operation of the Alteryx Server architecture described above:

  • 1 x Alteryx Server
  • 6 x Alteryx Server 2-Core Additional Capacity

Environment Summary

In this architecture, all of the four main service components of Server (Controller, Worker, MongoDB persistence, and the Gallery front-end) reside on a single node, and an additional Worker node is added into the environment. When we configure all four services on a single node, we generally recommend eight CPU cores as a starting point, which allows for four simultaneously processing workflows (using classic E1 engine). By including an additional Worker node, this allows a doubling in the number of workflows the environment can process simultaneously. Each node would be configured to allow four simultaneous job executions.

 

* Based on average behavior and recommended configurations. Your performance may vary.

 

 

Starter Architecture C

Three-node Environment

three.PNG

 

Specifications

  • Primary Node
    • Multiple services configured: Controller, MongoDB Database, and Worker
      • Worker configured to process QoS 6 (workflow validations)
    • 8 CPU Cores (physical)
    • 128GB RAM
    • Disk:
      • C : 100GB(+)
      • D : 500GB(+)
  • Worker node
    • 8 CPU Cores (physical)
    • 128GB RAM
    • Disk:
      • C : 100GB(+)
      • D : 1TB(+) SSD with installed Location/Business datasets, or 500GB(+) SSD without
    • Worker service configured
      • Worker configured to process four simultaneous workflows
  • Worker node
    • 8 CPU Cores (physical)
    • 128GB RAM
    • Disk:
      • C : 100GB(+)
      • D : 1TB(+) SSD with installed Location/Business datasets, or 500GB(+) SSD without
    • Worker service configured
      • Worker configured to process four simultaneous workflows

Capabilities 

  • Standard Server Functionality 
    • Ad-hoc, web-based workflow execution 
    • Scheduled workflow execution 
    • Centralized workflow version control 
    • REST API for execution, administration, and audit 
    • Centralized management of authorization (for execution, data connections, credentials) 
  • Capable of providing 480 minutes of workflow execution per hour * 
  • Capable of serving up to 100 concurrent web user sessions * 
  • Workload management options, including Worker assignment 
  • Redundancy for the Worker service 

 

Alteryx Licensing 

Alteryx Server uses a core-based licensing model. The following licenses will serve the operation of the Alteryx Server architecture described above: 

  • 1 x Alteryx Server 
  • 10 x Alteryx Server 2-Core Additional Capacity 

Environment Summary 

In this architecture, the Gallery and Worker processes have been offloaded from the Controller node, with the exception of a Worker process configured to only provide workflow validation (an operation that occurs when workflows are published). By including an additional Worker node, this allows an increase in the number of workflows the environment can process simultaneously. Each node would be configured to allow four simultaneous job executions, for an environment total of eight jobs  (using classic E1 engine). The Gallery service has also been offloaded from the Controller, which segregates the services more, allowing better access to resources. 

 

* Based on average behavior and recommended configurations. Your performance may vary.

 

Other Designs & Configurations 

The three starter architectures describe just the beginning of possible Alteryx Server architectures. These designs could be modified to suit specific needs. These starter designs aren’t intended to address more advanced considerations around full redundancy, high availability, and/or full service segregation. More advanced architectures will be discussed in subsequent articles in this series.  

 

Summary 

In this blog we have introduced the primary “starter architectures” often used when customers initially deploy Alteryx Server.  In subsequent blog entries in this series, we will look at a number of more detailed topics ranging from scalability, high availability, cloud deployments, and more.  If you have any topics you would specifically like to see discussed, please leave a comment below.  Thanks! 

Comments